Multi-signal analysis
Combine VRE, Sightengine, ExifTool, and C2PA evidence inside one report.
Start with 3 free checks and review explainable detector signals.
Detect whether an image was
generated by AI.
ImageVerity combines provenance, metadata, visual reasoning, and commercial detector scores so teams can make faster and more defensible image review decisions.
AI
Detection workbench
Upload a JPG, PNG, or WEBP image
Run EXIF review, C2PA provenance checks, detector scoring, and visual reasoning in one workflow.
# Newsrooms
# Commerce
# Moderation
# Trust & Safety
5
Review flow
1Prepare the assetReady
2Read EXIF metadataReady
3Check C2PA provenanceReady
4Run detector signalsReady
5Summarize reasoningReady
Explainable risk output
ImageVerity is designed to support operational decisions, not to claim absolute truth. High-risk or conflicting results should route to manual review.
That cautious framing is part of the product, not an afterthought.Why ImageVerity
A review workflow built for real production decisions
Instead of relying on one black-box score, ImageVerity helps teams look at provenance, metadata, detector outputs, and visual clues together.
Provenance-aware review
Recognize signed AI provenance when a trustworthy C2PA chain already confirms an origin.
Operationally cautious output
Use the result as a risk signal for moderation, newsroom, or commerce workflows instead of overclaiming certainty.
Use cases
Built for teams that need to review images at speed
From newsroom verification to seller onboarding and UGC moderation, the product is shaped around review queues and risk judgment.
News verification
Review suspicious visuals before publication and keep the reasoning visible to editors.
E-commerce risk control
Spot AI product composites, provenance gaps, and suspicious image reuse patterns.
Platform moderation
Add explainable image review signals into trust and safety workflows.
Latest writing
How image verification actually works in practice
Practical guides on telling AI images from real ones, reading provenance records, and knowing when a detector score should not decide the case.
Guide
How to tell if a photo is AI-generated: 7 checks anyone can run
Start with what you can see, then move to metadata and provenance. No single score decides it.
ExplainerDeepfake vs AI-generated image: what is actually different
Deepfakes target a specific person. AI-generated images are broader. The distinction changes which checks matter.
LimitsWhy screenshots destroy image evidence
A screenshot is a new image, not a copy. Here is what it removes and why detectors get it wrong in both directions.
ComparisonReverse image search vs AI detection: they answer different questions
One finds where an image was published. The other measures whether the file looks camera-native.
LimitsCan AI image detectors be wrong?
False positives and false negatives are routine. A single score should never be treated as final proof.
ProvenanceWhat C2PA can and cannot prove
A valid manifest tells you the file history. It does not tell you whether the depicted event happened.
See the evidence chain before you make the call.
Start with a small upload flow, review the report, and decide where manual escalation belongs in your workflow.